Instance Segmentation#

class model_api.tilers.instance_segmentation.InstanceSegmentationTiler(model, configuration={}, execution_mode='async', tile_classifier_model=None)#

Bases: DetectionTiler

Tiler for object instance segmentation models. This tiler expects model to output a list of SegmentedObject objects.

In addition, this tiler allows to use a tile classifier model, which predicts objectness score for each tile. Later, tiles can be filtered by this score.

Constructor for creating a semantic segmentation tiling pipeline

Parameters:
  • model – underlying model

  • configuration (dict) – it contains values for parameters accepted by specific

  • tiler (tile_size, tiles_overlap etc.)

  • execution_mode – Controls inference mode of the tiler (async or sync).

  • tile_classifier_model – an ImageModel, which has “tile_prob” output.

__call__(inputs)#

Applies full pipeline of tiling inference in one call.

Parameters:

inputs – raw input data, the data type is defined by underlying model wrapper

Returns:

  • postprocessed data in the format defined by underlying model wrapper